Early period: 1890-1930

During the last years of the nineteenth century and the first decades of this century, as social issues became topics of academic study, virtually all research was qualitative in nature. This is evident in early classic works by Weber, Durkheim, Simmel, and others (for references to key texts, see Berger and Berger, 1976: 26-55). As academic specializations were defined and university departments created, qualitative methods gained a solid foothold. Several factors may explain this emphasis. First, there was still a strong affiliation of social science with the mode of investigation utilized in philosophy and the humanities. Second, the social sciences were young and searching for global, overall perspectives; the essay format was more suitable to this task than that of the contemporary research article. Finally, what eventually became known as “the scientific method” had yet to be fully developed and applied to the social sciences.

The emphasis on qualitative research was especially evident in anthropology. Although methodological diversity in the discipline has since developed (see Sanday, 1983), the qualitative emphasis has continued to this day. The contributions of Malinowski (1922), Boas (1940), and Radcliffe-Brown (1952) were particularly influential in determining how field studies were conducted. Malinowski (1922: 25) is credited with encouraging first-hand observation in an effort “to grasp the native’s point of view, his relation to life, to realize his vision of his world,” even though, on later publication of Malinowski’s diaries, it appeared that he himself had difficulty putting this principle into practice (Malinowski, 1967).

In this early period, qualitative field research was introduced to sociology in the USA, and pioneering work was conducted at the University of Chicago. Founded in 1892, the sociology department there - then combined with anthropology - came to be known as the “Chicago School.” In the early decades of this century, it was to have major impact on the discipline. Under the influence of W.I. Thomas, Ernest Burgess, and Robert Park, a concerted academic enterprise developed around the study of urban life. Early Chicago studies concentrated on deviant groups in the city: hobos (Anderson, 1923), gangs (Thrasher, 1927), criminals (Sutherland, 1937). Other scholars at the university introduced community studies, later known as urban ethnography. Polish immigrants to the city were the subject of a momumental study (Thomas and Znaniecki, 1918-20), and the structure of small-town community life was also explored (Lynd and Lynd, 1929). A third, later cluster of studies examined professional life - the police (Westley, 1951), businessmen (Dalton, 1959), teachers (Becker, 1951), and doctors (Hall, 1944).

The Chicago School was the home of a long list of prolific sociologists, but the contributions of one stand out: Robert E. Park. Having worked for some time as a journalist, he retained interest in the media as institutions within society. In a collection of essays on the city (Park et al., 1925), Park contributed a piece entitled “The natural history of the newspaper” (reprinted in Schramm, 1960). Although without empirical data supporting his interpretation of the place of print media in a historical process, the piece does exemplify Park’s concern for newspapers in the context of city and community. This concern was also evident in his study of the immigrant press (Park, 1922) in which he examined the function of newspapers among European immigrants.

Frequent reference has been made to Park’s recommendation that social scientists imitate the work routine of newspaper reporters. His instruction to a student about to begin a research project was, “Write down what you see and hear; you know, like a newspaper reporter” (quoted in Kirk and Miller, 1986: 40). Implicit in this recommendation is the assumption that the discernment of “facts” is unproblematic, and that facts can be gathered and analysed straightforwardly. This position, not surprisingly, has since been challenged and modified, emphasis now being placed on the social construction of “facts” (see van Maanen, 1988: 18).

Another bit of advice Park is said to have stressed was to leave the protected confines of the university and to explore the city:

Go and sit in the lounges of the luxury hotels and on the doorsteps of the flophouses; sit on the Gold Coast settees and on the slum shakedowns; sit in the Orchestra Hall and in the Star and Garter Burlesk. In short, gentlemen, go get the seats of your pants dirty in real research.

(quoted in McKinney, 1966: 71) Park’s emphasis here on first-hand observation is often taken as evidence that the Chicago School practiced what is currently known as participant observation. Even though many Chicago sociologists became intimately familiar with the cultures studied, they seldom participated in those cultures as researchers. Most early Chicago studies relied on document analysis. In only two studies - Cressy’s (1932) on dance halls and Anderson’s (1923) on hobos - is there explicit reference to participant observation as a method of data collection, and even then there is limited explication of how the method was employed (Hammersley, 1989: 1-84). Harvey (1987: 50) suggests that it is misleading to consider these studies forms of participant observation as the term is now employed. (For another perspective on the place of participant observation in Chicago School research, see Chapter 3, note 2, by Tuchman in this book.)

Members of the Chicago School also conducted communication research on the effects of films on children. In the late 1920s, Herbert Blumer and Philip Hauser were commissioned by the Motion Picture Research Council to investigate the relationship between film and delinquency. As part of the so-called Payne Fund Studies (the agency which financed the research for the Council), two monographs were produced (Blumer, 1933; Blumer and Hauser, 1933; see also Blumer, 1935, reprinted in Short, 1971). This work exemplifies several aspects of the Chicago heritage. Both research projects were intended to “capture the attitudes or perspectives which mediate the effects of objective factors, in this case of films” (Hammersley, 1989: 89). The studies were exploratory in nature, with minimal methodological explication. There was heavy use of interviews and life histories, and little attention was given to data collection through participant observation.

The debate on social-scientific methodologies in the USA in the 1930s and 1940s also came to include a number of immigrants from Europe, representing different theoretical and political orientations. One group included refugees from the Frankfurt School (see Jay, 1973); their approaches comprised particularly qualitative textual analysis and historical studies (see Chapter 6 by Larsen in this volume). Another orientation was later to be developed into a mainstream of American communication research methodology by Paul F. Lazarsfeld and his collaborators, thus superseding the positions both of the Chicago School and of critical theory.

Indeed, during the period of these debates and studies, the application of quantitative methods increased. As others have noted (Bulmer, 1984; Harvey, 1987), there was never antipathy toward statistics at Chicago; quantitative methods were often employed alongside qualitative ones. But, by the 1930s a separate division with a quantitative orientation had developed at Chicago, and the methodological debate then emerging across the nation began to take explicit and emotional form.

Middle period: 1930-60

The methodological debate had its roots in the rise of positivism. Since the late 1920s and early 1930s, sociologists had begun to turn to methods of research as practiced in the natural sciences, or, to be precise, as natural science was perceived from the perspective of social science at this historical moment. It is perhaps an irony of history that social scientists were seeking to construct an objectivist notion of social reality at a time when most other arts and sciences, including physics, were arriving at a multiperspectival conception of the reality under inquiry (Lowe, 1982:109-17). Researchers both in and outside Chicago were developing quantitative measuring devices and conceptual schemes that were intended to elevate the status of sociology to a science. The model for empirical research eventually became the one perceived as the standard in physical sciences, particularly physics. Experimental designs came to dominate research thinking, along with hypothetico-deductive reasoning, and methodologies emphasized the use of “objective” data-collection techniques and the standardization of analytical procedures.

In the course of the 1930s, then, proponents of quantitative methods gained the upper hand in the methodological struggle. The Chicago-style case study had all but disappeared as a mode of social science by the 1950s. Survey research had become the method in the social sciences; as Benney and Hughes (1956: 137) remarked, modern sociology had become “the science of the interview” (article reprinted in Denzin, 1970b). As the influence of positivist theory and quantitative methodology reached its peak in the 1950s, qualitative research came to be seen as a preliminary activity which could, at best, lay the groundwork for “real” science.

One of the Chicago studies of this period, expressing the tension between the qualitative and quantitative traditions, focused on the community press (Janowitz, 1952). Although clearly rooted in the early Chicago interest in urbanism and qualitative approaches, the study also systematically employed quantitative research methods. An appendix to the study elaborates on coding procedures for a content analysis, and the material presented about a survey (sampling information and instrument design) is similar to that found in most contemporary survey research monographs.

Several factors contributed to the demise of qualitative methodologies such as the Chicago case-study approach. First, there was a desire to create a genuine “science” of social investigation, modeled on positivism and the successes of the physical sciences. Second, fueled by World War II, calls were made for research to measure the impact of communication, in particular propaganda. These concerns were accompanied by a surge of funding for scientific findings. It was during this period that the Bureau of Applied Social Research at Columbia University established itself as one of the main centers of mass communication research in the USA. Last but not least, after the war there appeared to develop a structural need for social-scientific knowledge which could be applied to the development of industry and to the planning of social and educational institutions (Galbraith, 1967). Such social engineering, as witnessed by policies in many Western countries, found a theoretical ally in the functionalist perspective then taking hold. Both social policy practice and functionalist theory were well served by the survey research methodology then coming of age.

Late period: 1960-present

After functionalism and quantitative methodologies had held sway through the 1950s and into the 1960s, the theoretical and political critique of these positions intensified in the 1960s. Gouldner (1970), for example, identified the “coming crisis” of sociology as a disenchantment with functionalism and “grand theory.” Where, a decade earlier, the discipline had still been preoccupied with building and securing its institutions, the movement now was toward an application of the discipline to the deep social conflicts outside academia. Rooted in political, economic, and racial inequalities, these conflicts mobilized students, blacks, other minorities, and, sometimes, social scientists (Colfax and Roach, 1971), not just to study events, but to become actively involved in organizations and demonstrations. For economic and social-structural reasons, of course, such involvement had become more feasible by the 1960s.

This societal context provided the backdrop for disillusionment with positivist versions of social science also in the academic institutions. One point of frequent criticism referred to the limitation of these approaches when studying human behavior and especially its origins in a social reality as experienced and lived. Another criticism singled out some sociologists’ apparent obsession with scientific method. Herbert Blumer, former staff member of the Chicago School, became one of the most vocal critics of the increasing emphasis on quantitative methods in sociological research, an effort characterized in one piece (Blumer, 1954) as a tendency to reduce social existence to variables.

Outside Chicago, C. Wright Mills, then at Columbia University, was also a vehement opponent of mainstream sociology. His Sociological Imagination (1959) was a landmark critique of Parson’s efforts to construct “grand theory” and the empiricism then dominant in sociology. In apparent prescience of Feyerabend’s (1975) attack on traditional models of science, Mills urged all researchers to become their own methodologist (Mills, 1959: 121). Many other scholars (see Phillips, 1971; 1973) systematically listed the shortcomings of survey research and argued for a modification of standardized research procedures.

In the field of communication research, equally, criticism of predominant research methodologies was expressed. Gitlin (1978) summarized much of the reservation felt for the dominant method of mass communication research, the survey. Though Gitlin did not explicitly call for a qualitative approach to media research, he did argue that the minimal effect of communications found through surveys was largely a product of the methodology.

In a recent historical sketch of communication research, Dennis (1988) observed that media studies were conforming to a methodological metamorphosis of the social sciences generally. By the late 1970s, the supremacy of quantitative research had been “challenged by qualitative researchers - many with an ideological bent -who decried quantification and questioned the utility and value of the prevailing research tradition” (Dennis, 1988: 4). Interpretive forms of social inquiry have played a central role in this transformation.

Symbolic interactionism

Symbolic interactionism was the form of interpretive inquiry at the center of the theoretical and methodological reorientation of the 1960s and 1970s. It is grounded primarily in Mead’s (1934) Mind, Self, and Society, described as the “single most influential book, to date, on symbolic interactionism” (Manis and Meltzer, 1967: 140). Others - Cooley (1930), Thomas (1928), and Dewey (1925) - also contributed to its development, but there is general agreement that the refinement of the theoretical position came from Blumer (1969). He posited, first of all, that people act on the basis of the meaning they themselves ascribe to objects and situations. Second, Blumer held that meaning is derived from interaction with others, and that this meaning is transformed further through a process of interpretation during interaction (Meltzer etal., 1975: 2). Coupled to these notions was a methodology stressing respect for the world and actions of individuals as well as non-intervention by the researcher in that world, what is often referred to as a naturalistic perspective (Lincoln and Guba, 1985). Specifically, participant observation is normally associated with the naturalistic perspective and generally with the work of symbolic interactionists (Ackroyd and Hughes, 1981: 102-3; Rock, 1979: 178).

Other sociologists also contributed to the development of the naturalistic perspective. Goffman (1959), for example, is credited with creation of a distinct “dialect” of symbolic interactionism, the dramaturgical approach (see Meltzer et al., 1975). Some communication scholars (Tuchman, 1978) have been influenced by one of Goffman’s (1974) concepts, framing, which he himself applied to the media in his book Gender Advertisements (1976).

Even while symbolic interactionism is correctly considered the motor behind recent developments in qualitative methodology, it would be a mistake to assume that the two terms are synonymous. In the first place, there is a group of symbolic interactionists who conduct research from a positivist perspective. The Iowa School of symbolic interactionism has employed survey research and standardized observation techniques in an effort to operationalize the concept of self (Meltzer et al., 1975: 55-9). In the second place, other varieties of interpretive inquiry have made significant contributions to qualitative methodology.

Ethnomethodology

This second form of interpretive inquiry seeks to identify the rules people apply in order to make sense of their world. Whereas ethnomethodology originates from the work of European phenomenologists, in particular Schutz (1967), the central figure in its later development has been Harold Garfinkel, who conceived the term, formulated the core ideas, and has served as a source of inspiration for other ethnomethodological researchers. For Garfinkel, ethnomethodology is a form of “practical sociological analysis” (1967:1). This sociological analysis, however, is not merely an undertaking of professional sociologists: ethnomethodology is an everyday activity in which social agents constantly engage as they arrive at an interpretive understanding of other agents and actions through interaction, thus making sense of social reality. Media studies that draw, in part, on this approach include Molotch and Lester’s (1974) examination of news as purposive behavior and Tuchman’s (1978) investigation of news organizations.

There is no one research method associated with ethnomethodology. Nevertheless, participant observation and in-depth interviewing are frequently employed as elements of an open research strategy. A prominent place is almost always given to everyday conversation, this being the primary medium of everyday interaction. A special approach, refined to perfection by Garfinkel and his colleagues, is that of experiments which are designed to disrupt taken-for-granted rules of conversation. As Garfinkel (1967: 37) himself described the technique, “Procedurally it is my preference to start with familiar scenes and ask what can be done to make trouble.” Douglas (1976) in particular has contributed to the development of this approach, and has in the process ignited a continuing debate on the ethical legitimacy of such research strategies (Cavan, 1978). The justification given for using a disruptive means of research is its end result: the underlying rules governing behavior in everyday situations.

A landmark study of interpretive social inquiry into media, drawing according to its references on both symbolic interactionism, ethnomethodology, and ethnography, was Lull’s (1980) examination of the social uses of television. He asked how media, particularly television, “play a central role in the methods which families and other social units employ to interact normatively” (Lull, 1980: 198). Using a combination of participant observation and interview research methods, Lull and his associates examined interaction and communication patterns in their natural setting - the home. This work prepared the way for the many recent studies in the ethnography of mass communication.

Ethnography

Ethnography stems from anthropology; indeed, there was a time when the two concepts were considered identical (Kuper, 1973:14). Today, ethnography is practiced in other disciplines, and also within anthropology several versions of ethnography now exist (Sanday, 1983; see also Clifford and Marcus, 1986; Ellen, 1984). Despite such diversity, most anthropologists seem to agree on three core principles. First, ethnographic research is concerned with cultural forms in the widest sense of the term, including the everyday as well as religion and arts (see Fetterman, 1989; Hammersley and Atkinson, 1983). Second, studies generally acknowledge the need for long-term participant observation, with the researcher serving as the primary instrument of inquiry. Finally, multiple data-collection methods are generally employed, according to Sanday (1983: 21), as a check on observational findings. (For criticism of this technique, see the discussion of “triangulation” later in this chapter.)

Sanday (1983) identifies three types of anthropological ethnography: holistic, semiotic, and behavioristic. Of these, the holistic variety has the longest tradition and is dominant within the discipline. “Holistic” refers to the scope of inquiry and specifically to the purpose of investigating many aspects of the particular group or society being studied. For this reason, anthropological ethnographies have commonly been situated in clearly defined settings such as a village or other small geographical community. Finally, the establishment of rapport between the researcher and the group studied is considered a critical element of ethnography (Seiter etal., 1989b). This is one of the reasons why some researchers (Wolcott, 1975) suggest a minimum of one year for the fieldwork phase of an ethnographic project.

An increasing number of communication studies claim to be conducted in the ethnographic tradition. Lull (1988a) has edited a volume of such work which concentrates on family television viewing in different cultures. However, one problem with some of the research which is labeled as ethnographic is that it is unclear what the studies have in common with either the research tradition or its methodology. Lull himself has complained, in a critique of the cultural studies approach to media audiences, that “ethnography has become an abused buzzword in our field” (Lull, 1988b: 242).

Another critic (Braber, 1989) examined three ethnographic studies of women and popular culture, and came to essentially the same conclusion - that much of what passes as ethnography deviates considerably from what at least anthropologists mean by the term. Braber examined Hobson’s (1982) analysis of the British soap opera Crossroads, Radway’s (1984) study of female readers of popular romance fiction, and Seiter and colleagues’ (1989b) research on soap-opera viewers. All three studies were limited, to varying degrees, in regard to the aspects of anthropological ethnography that Braber considered: centrality of the concept of culture, employment of participant observation, and smallness of the research setting.

Seiter’s research team, aware of these difficulties, were frank in discussing the methodological shortcomings of their work: ethnographic research of television audiences, they conceded, “have not satisfied the requirements of ethnography proper, and our study is no exception” (Seiter etal., 1989b: 227). Among the problems they noted in ethnographic media studies were limitations on the frequency and duration of the contact with informants and the resulting difficulty of establishing rapport. Some of these studies (Katz and Liebes, 1984; Lull, 1980) also tend to consider only a single aspect of a culture - television programming - and hence cannot claim a holistic approach.

In spite of these problems, the call for more ethnographies of media audiences is frequent and forceful. Ien Ang, in her book Desperately Seeking the Audience (1991), disparages the emphasis on decontextualized quantitative data, largely promoted by the television industry, and recommends “ethnographic understanding” of audiences as an alternative (Ang, 1991). Still, although she argues at length for the value of ethnography in audience research, she devotes no more than a footnote to the characteristics of ethnography as a concrete, empirical approach to conducting research.

The challenge remains, then, to explicate adequately the principles and procedures of ethnography in order to avoid its identification with any and all qualitative methods. This task is particularly urgent because of the potential of ethnography as a form of interpretive inquiry for mass communication research. For all the discussion and disagreement among anthropologists about the nature of ethnography, the anthropological consensus, as noted above, may provide the best starting point for designing ethnographic media research. At the same time, there are a number of general issues outstanding in the scientific debate on how to develop systematic and applicable qualitative research projects.

Issues of interpretive inquiry: theory and politics

Interpretive sociology generally has been the focus of much criticism, directed in particular at the non-political stance of the work. Its advocates have been accused of failing to take into account institutional power and structural determinants as limitations on the individual’s freedom of action (McNall and Johnson, 1975). This critique, especially with respect to the apolitical and relativistic stance, applies to much of the work of ethnomethodologists, but less so to that of symbolic interactionists. One of the major proponents of symbolic interactionism, Howard Becker, has argued that it is not possible “to do research uncontaminated by personal and political sympathies … and that the question is not whether we should take sides, since we inevitably will, but rather whose side we are on” (Becker, 1967: 239). The general issue of political commitments and knowledge-interests of research has remained on the agenda of interpretive inquiry, and feminist researchers have made significant contributions to the debate.

Also within feminist scholarship there is considerable disagreement regarding appropriate methodological approaches to social science. Steeves (1987) suggests that feminists with a critical studies orientation tend to employ qualitative research methods, while those with a traditional social science background often use quantitative ones. Radical feminists generally tend to dismiss quantitative methods as “masculine” strategies of knowledge and to prefer qualitative strategies such as in-depth interviewing and participant observation. Interestingly, feminist researchers, wishing to document and highlight the everyday life of women, find methodological support in interpretive forms of inquiry from open interviews to life histories (see Roberts, 1981).

Some of the crucial issues addressed in feminist research concern the relationship between the researcher and the subject of study. One question is whether and to what degree the researcher should maintain a distance from the researched, who are frequently other women with whom the (female) researcher may empathize (Oakley, 1981). Another issue is the legitimacy of an emancipatory or action component in the research strategy, the purpose being to change an inequitable state of affairs (Mies, 1979).

Both of these issues - the politics and the epistemology of research - are interrelated and raise a fundamental question: what is the relationship between the objective of a study, on the one hand, and the objectivity of the research procedures and findings, on the other? This question, to be sure, is not unique to feminist research. In particular, the action research tradition, relying in part on qualitative methodology, also has its roots in an emancipatory objective of social science. The work of Negt (1968) and Freire (1974), stressing the involvement and mobilization of the researched, has informed participatory research, a variant of action research. Participatory research specifically has been applied to mass communication, both in communication development projects (Camilo etal., 1990; Coesmans and van den Goor, 1990) and in studies of local radio stations in Latin America.

To sum up, there are affinities between the qualitative tradition and research with an emancipatory objective. However, it is incorrect to assume that most early or current qualitative research is inspired primarily by such motives. While it is true that some figures associated with the Chicago School were guided by progressive ideals (see, for example, Dewey, 1927), its research program was not designed to solve social problems. Only the Chicago sociologist Burgess engaged in research projects that grew out of his own social and political involvement, and even then he was mainly interested in basic research, and only in the second instance in policy-oriented studies (Harvey, 1987: 36-7).

Before and after the Chicago School, of course, the personal responsibility of researchers for the political implications of their work has been a contested issue in different scientific fields, and is likely to remain so in the future. We submit that Becker’s position, that we ultimately “choose sides,” simplifies the matter. Few social and political issues can be reduced to categories of “underdogs” and “suppressors,” allowing the social scientist an indignant rejection of the latter. Gouldner (1968), in debate with Becker, argued for choosing sociology. We would endorse that position, further recommending, as an aspect of sociology or of media research, explicit analysis of the researcher’s own social and political stance, both publicly and in the academic forum. (For a discussion of such public debate on mass communication research as a form of meta-communication, see Chapter 12 in this volume.) In the process, qualitative research may gain in its relevance and legitimacy.

Data collection

Data collection in qualitative research involves a variety of techniques: in-depth interviewing, document analysis, and unstructured observations. Though these techniques are often referred to by a single term - participant observation - this is in fact misleading. Furthermore, in a number of cases it is incorrect to associate qualitative social science with participant observation. Many qualitative studies, for example the early Chicago studies, rely on a single data-collection method, either document analysis or interviewing.

Researchers have frequently asserted that a particular method was superior to all others. Thomas and Znaniecki made such a claim regarding life histories: “We are safe in saying that personal life-records, as complete as possible, constitute the perfect type of sociological material” (quoted in Madge, 1962: 61; emphasis in original). Similar assertions were made in a debate between Becker and Geer (1957; 1958), on the one hand, and Trow (1957) on the other (exchange reprinted in Filstead, 1970), the former arguing for the virtues of participant observation and the latter for interviewing.

Such declarations, we suggest, are presumptuous; Bulmer (1984: xv) has rightly pointed to the unproductiveness of engaging in a “best method” debate in an absolute sense. In fact, in Becker and Geer’s rejoinder to Trow they agree with him on the point that the problem under investigation dictates the method to use. This point is illustrated further in a typology constructed by Zelditch (1970) and refined by Denzin (1970a: 30-1), in which techniques of data collection are correlated with types of information (Table 2.1).

The typology suggests that participant observation is best suited for case studies and life histories (“incidents and histories” in Table 2.1), and least suited for overviews of entire populations (“frequency distribution”). For the study of organizations (“institutionalized norms and statuses”) participant observation is deemed an adequate, but inefficient data-collection method. Interviewing, according to Zelditch, seems a viable data-collection device for all three types of studies, but less appropriate for surveys of large groups or populations (“frequency distributions”) than for the study of cases (“incidents”) and organizations (“institutionalized norms and statuses”). What Zelditch calls enumerations refers to survey research methodology and is considered most appropriate for the study of the distribution of characteristics in a population.

There are, of course, limitations to such a typology, and they become clear when one attempts to place specific qualitative communication studies in the cells of the table. Participant observation, for example, was the primary data-collection method employed by Gans (1979) in his study of news organizations; his results were more than “adequate.” Interviewing, on the other hand, has been employed in a wide variety of studies of media organizations and institutional procedures, but, we suggest, it is far from always “efficient” or the “best form.” As Deutscher (1973) demonstrated, there is often a discrepancy between reports of attitudes gained through interviews and observations of the behavior related to those attitudes. Once again, the “how” of research (methodology) should be deliberated carefully in each particular case with reference to “what” and “why” (the subject matter and purpose of inquiry).

In order to examine the explanatory value of specific methods, we next consider more closely participant observation, which is frequently identified as the ideal method for qualitative research. Second, we discuss the relevance of employing multiple methods, what is commonly known as “triangulation.”

An often quoted definition of participant-observation has been offered by Becker and Geer (1957: 28):

By participation observation we mean that method in which the observer participates in the daily life of the people under study, either openly in the role of researcher or covertly in some disguised role, observing things that happen, listening to what is said, and questioning people, over some length of time.

The primary purpose of participant-observation research, accordingly, is to describe in fundamental terms various events, situations, and actions that occur in a particular social setting. This is done through the development of case studies of social phenomena, normally employing a combination of data-collection techniques. Other definitions further stress the multiple methods of participant observation:

it is probably misleading to regard participant observation as a single method … it refers to a characteristic blend or combination of methods and techniques that is employed in studying certain types of subject matter: primitive societies, deviant subcultures, complex organizations … social movements and informal groups … [it] involves some amount of genuinely social interaction in the field with the subject of study, some direct observation of relevant events, some formal and a great deal of informal interviewing, some systematic counting, some collection of documents and artifacts, and openendness in the direction the study takes.

(McCall and Simmons, 1969: 1) One problem with this latter definition is that it includes nearly every form of data-collection and interpretation under the heading of participant observation. Hence it becomes difficult to discriminate and, most important, to compare and assess the findings that different data-collection methods generate. Participant observation is best suited, in comparison with survey or experimental designs, for interpretive inquiry into social interaction from the perspective of the people involved.

In previous research, the multiple method approach is best known under the term triangulation, and has been advocated most vocally by Webb and colleagues and, later, by Denzin:

If no single measurement class is perfect, neither is any scientifically useless … for the most fertile research for validity comes from a combined series of different measures, each with its idiosyncratic weakness, each pointed to a single hypothesis.

(Webb etal., 1966: 174)

Triangulation, or the use of multiple methods, is a plan of action that will raise sociologists above the personalistic biases that stem from single methodologies.

(Denzin, 1970b: 27) One of the assumptions of a multiple method strategy is that such an approach provides for more valid results than a single research strategy. Or, as Jick (1979: 604) puts it, the basic assumption of all triangulation is “that the weaknesses in each single method will be compensated by the counter-balancing strengths of another.”

Various forms of triangulation have been proposed (for example, Brewer and Hunter, 1989). One of the most elaborate developments of the technique includes four types: triangulation of the data, the investigator, the theory, and the method (Denzin, 1970b). Data triangulation refers to the dimensions of time, space, and analytical level in which information is obtained. Investigator triangulation involves the more standard approach of using several analysts or coders, often as part of a multidisciplinary team of scientists. Theoretical triangulation suggests application of concepts and perspectives from diverse theories and disciplines. Finally, methodological triangulation constitutes a research strategy in which different methods are employed for data gathering and analysis around a single object of study.

Some calls for triangulation may be rooted in a scientifically naive notion that multiple methods can reveal a single, “true” reality beyond frameworks of theory and interpretation. Phillips (1973: 91), for one, has raised the question of whether triangulation actually increases rather than reduces biases inherent in particular data-collection methods. In spite of such reservations, triangulation may become a constructive force in the development of methodology as well as theory. For one thing, it can stimulate inventive uses of familiar research methods, and thus may help to uncover unexpected dimensions of the area of inquiry. For another, given appropriate theoretical and meta-theoretical reflection on the status of each set of data and findings, it may at times allow for more confidence in the conclusions of qualitative studies. Perhaps most important, triangulation can assist in constructing a more encompassing perspective on specific analyses, what anthropologists call “holistic work” or “thick description” (Jick, 1979: 608-9; also Geertz, 1973).

What should be noted, finally, is that triangulation does not absolve qualitative researchers of interpretive work. Indeed, when findings derived from different methods conflict or fail to corroborate each other (as well as when they support each other), this signals not the end of the study, but the beginning of a phase of theoretical analysis examining the nature of agreements and disagreements. To repeat, further empirical but also theoretical work is needed to specify the explanatory value of different methods of data collection. This may become a priority of qualitative research, along with the development of systematic analytical procedures.

Participant observation, often including triangulation, has been applied also to processes of mass communication. In particular, studies of media organizations have been a proving ground for this data-collection method. Exemplary works include Gans’ (1979) study, Deciding What’s News and Tuchman’s (1978) Making News. At the audience end of the process, Lull (1980), as already noted, has conducted pioneering work on television audiences based on participant observation, which has inspired a new generation of audience researchers committed to ethnographies of media use in the natural setting of the home (see Lull, 1988b). However, it is striking that in these media studies, as in much research from other disciplines, there is little or no indication of how the collected data were analysed.

Analysing qualitative data

Early writers in social science who attempted to convey the craft of qualitative analysis to other scholars, seldom went beyond recommending filing systems for documents, wide margins on field notes to pen in codes, and carbon copies of all documents to allow later cutting and pasting of the material. The lack of sophistication of these aids and procedures has not passed unnoticed by critics of the qualitative approach. Also researchers who recognize the relevance and legitimacy of qualitative research have reservations about the systematicity of current approaches, as summed up in the following question: “How can we be sure that an ‘earthy,’ ‘undeniable,’ ‘serendipitous’ finding is not, in fact, wrong?” (from Miles, 1979; quoted in Miles and HÜber man, 1984: 16). Above all, there is a call for explicit and systematic procedures of analysing qualitative data sets. In the words of one sympathetic critic over a decade ago, there was, at that time, a strong need for “systematic methods for drawing conclusions and for testing them carefully - methods that can be used for replication by other researchers, just as correlations and significance tests can by quantitative researchers” (Miles and Huberman, 1984: 16). It would be an exaggeration to suggest that fully developed and systematic methods of analysis are now available, but developments have since occurred both in aids to organizing and conducting analyses, and in analytical procedures.

Aids to analysis

Among the basic aids to analysis are the various primers on the mechanics and procedures of fieldwork already noted (Johnson, 1975; Schatzman and Strauss, 1971; Wax, 1971), suggesting practical approaches and rules of thumb for various purposes of qualitative analysis. An important recent contribution to the literature is Miles and HÜber man’s (1984) volume Qualitative Data Analysis: a Sourcebook of New Methods, which provides suggestions and examples of how to systematize three aspects of analysis: data reduction, data display, and the drawing of conclusions. Data reduction refers to the processes of selecting, distilling, and otherwise transforming the information - data - found in field notes or interview protocols. Data display refers to various methods for visually rearranging data in the form of matrices, graphs, and charts. The suggestions of Miles and HÜber man represent additions and, in some cases, advances in relation to the aids already available for performing qualitative analysis. It should be noted, however, that “the drawing of conclusions” cannot be reduced to such aids, but centrally involves the researcher as an agent of analysis and interpretation.

One noteworthy aid of recent vintage is the computer. The application of computers to qualitative data analysis has lagged considerably behind their uses in quantitative research, in part because of prejudice among qualitative scholars. According to Denzin (1970a: xi), computer analysis and other high-speed data-processing techniques “too often become substitutes for the sociological imagination.” Such reservations have diminished with time, and many efforts have gone toward using computers for performing aspects of qualitative analysis, in particular various time-consuming administrative chores. A special issue of the journal Qualitative Sociology on computers in qualitative research was published in 1984. Since then, several other authors (de Graauw et al., 1986; Peters and Wester, 1988; Pfaffenberger, 1988; Tesch, 1990) have devoted attention to this topic.

A review of the various tasks which can be performed by computer is offered by Peters and Wester (1988). First, the machines can store and retrieve material better than any mechanical system. Through elementary editing or word-processing programs it becomes possible to type interview or observation protocols into computer files that can be printed on demand. Second, such files can be segmented, organized, and reorganized once codes have been assigned to their elements, thus dispensing with laborious and personalized cut-and-paste schemes. Various computer programs - “The ethnograph” (Seidel and Clark, 1984), “Qualog” (Hiemstra et al., 1987), and “Kwalitan” (Peters and Wester, 1990) - provide the possibility of coding data and then sorting the material through combinations of codes. Finally, computer files of qualitative data sets allow other researchers to gain access to the materials and hence to conduct further analysis. This development is likely to contribute to the perceived applicability and legitimacy of qualitative analysis among the community of media researchers.

One major use of the computer in qualitative research is for analyses of data sets through repeated iterative or cyclic procedures. Analysis of the “raw” data takes place continuously throughout the qualitative research process, even while the nature and intensity of the analysis may change depending on the specific stage of the study (Lofland, 1971: 117-18). The relevance of cyclic analysis stems from a central characteristic of much qualitative research whose purpose is theory construction - the exploration and refinement of concepts during the course of the study itself. The computer facilitates repeated coding and recoding of the basic data as theoretical notions and concepts are reformulated and developed.

For all of the potential of a computer, it is important to stress that the machine is doing no more than facilitating or aiding the process of analysis. The core value of the computer in qualitative research is accuracy and speed in the organization and administration of data. For example, in the selection of materials to illustrate a particular point or to develop a typology, there is much less opportunity to miss relevant material, which may frequently happen when the sorting of data is done by hand. Ultimately, however, the agent of analysis at each step of the qualitative research process is the researcher, not the machine (Peters and Wester, 1988: 337).

Analytical procedures

These aids to analysis, then, should be seen in relation to systematic procedures for performing qualitative analysis. Two analytical procedures - analytic induction and grounded theory - are of special interest here because of their historical origins and substantive contributions within social science. (See also the account in Chapter 1 of discourse analysis as a possible “statistics” or systematics of qualitative analysis.) Other procedures, such as ethnographic (Spradley, 1979; 1980) and phenomenological analysis (Hycner, 1985), represent refinements specifying the steps to follow in concrete analysis, but they are essentially varieties of the other, established procedures, and are not considered further in this context.

Analytic induction. Perhaps the earliest explication of a procedure for qualitative analysis was analytic induction, which involves “an exhaustive examination of cases in order to prove universal, causal generalizations” (Manning, 1982: 280). The procedure has been most elaborately worked out in a twelve-step sequence by Denzin (1970a; 1978). Basically, the procedure calls for, first, constructing a general description of the phenomenon under study. Next, the characteristics which the researcher initially assumes are the most important are elaborated and specified. Then, a specific case is examined to establish whether the assumed characteristics apply. If the case does not fit the characteristics, then either the description is modified so as to manifestly exclude the case, or the originally hypothesized characteristics are changed so that the case may become part of the phenomenon under study. This procedure is repeated until there are no more cases left to categorize, or until no cases arise which do not fit within the parameters of the phenomenon.

As a practical form of analysis, analytic induction is time-consuming and has found limited application outside exploratory sociological studies (see Lindesmith, 1947). It is, moreover, doubtful whether it lives up to the claim of being able to predict events, establish causality, or produce universal statements (Manning, 1982: 294). The technique does, however, offer a procedure for thoroughly examining cases that might be related to a concept in development. One example of work in communication research that relies on analytic induction can be found in the Lang and Lang (1953) investigation of the differences between the television coverage of the MacArthur Day Parade in Chicago and the perceptions of the event among the spectators along the parade route (see Chapter 11 of this volume).

Grounded theory. The procedure of analytic induction provided part of the inspiration for other researchers who were concerned with theory development while, at the same time, wanting analyses to remain “close to the data.” Glaser and Strauss (1967), two of these researchers, proposed that new theoretical formulations were needed which would be based or “grounded” in empirical data. They recommended relying on “sensitizing concepts” to guide such theory development, a phrase originally coined by Blumer:

Hundreds of our concepts - like culture, institution, social structure, mores, and personality - are not definitive concepts but are sensitizing in nature. They lack precise bench marks which allow clear-cut identification of a specific instance, and of its content. Instead they rest on a general sense of what is relevant.

(Blumer, quoted in Rock, 1979: 9) One of the difficulties with this proposal stems from the principle of staying close to the data. The question is how close to the data one can be and still undertake theoretical work, which of necessity requires a certain level of abstraction and hence distance from empirical data. Addressing this question, Glaser and Strauss (1967: 3) speak in terms of “criteria of fit” and “criteria of relevance.” Relevance, for them, not only has an analytical dimension, as in traditional deductive use of theory, but also a substantive dimension. Theories thus should “fit” the specific field under observation; the theoretical concepts are in this way “sensitized” to the subject of research.

While the analytical procedure proposed by Glaser and Strauss received widespread attention, it was also strongly criticized for being a polemic rather than a constructive intervention into scientific debate. As it turned out, researchers who set out to practice the precepts of grounded theory frequently went aground in uncharted analytical terrain. Glaser (1978), Strauss (1987), and Strauss and Corbin (1990) have made attempts to solve such difficulties in subsequent volumes, and other researchers, such as Turner (1981), have contributed to a further codification of analysis within grounded theory.

One of the most comprehensive efforts so far in this area is Wester’s (1984; 1987) procedural approach to grounded theory. This approach is comparable, in certain respects, to analytic induction, consisting of four phases which each in turn contain some fifteen procedural steps. The initial, or exploratory, phase is intended to extract preliminary concepts from the collected material. In the second or defining phase, the researcher tries to construct variables based on the concepts. In the third or reduction phase, the aim is to formulate the core of a theory. In the fourth and final phase, termed integration, the concepts are related to one another and the relations tested on the data. The cycle of reflection, observation, and analysis is repeated throughout the research process in each of the four phases until the theoretical formulations have exhausted the available data (Peters and Wester, 1990).

As was the case for analytic induction, communication research seldom makes explicit use of grounded theory. Lull (1988a: 16) does refer, albeit briefly, to the “compelling argument of Glaser and Strauss,” and he further explains his own preference for the grounded theory approach thus:

the theoretical essence of our work emerges quite spontaneously within e^ch research project. I believe that we should not simply conduct research that is programmatically influenced by any fixed theoretical perspective if we are to really “let the data speak to us.”

(Lull, 1988a: 17) Data of and by themselves, however, cannot generate theory. It is only through intervention by a researcher, operating within a theoretical perspective, that data can be examined and used to develop theory. For this reason, many researchers employ “sensitizing concepts” or ideal types in the preliminary phases of their empirical investigations. Such concepts may help to orient the researcher theoretically, while at the same time allowing the kind of flexibility which Lull was referring to. Most of the chapters in this book provide examples of the specific relevance of qualitative methodologies for theory development (see especially Chapter 11).

Reporting qualitative research

The final step of the qualitative research process - reporting the work - is sometimes overlooked, but deserves mention in this context since it is the point of contact with other researchers as well as with the interested lay public. Some authors, in fact, suggest that it is during the writing up of qualitative research that the final analysis of the data takes place (Maso, 1987:118; Miles and HÜberman, 1984: 213).

According to Burgess (1984: 182) there are three forms of qualitative research reports: (1) descriptions which make little or no reference to theoretical perspectives; (2) analytical discussions based on concepts emerging from the study; and (3) substantive accounts intended to contribute to general theory. Other scholars have attempted to discern the essential nature of the qualitative research report. Lofland (1971: 5) proposed that, (a) the report should get “close to the data” and should be based on a relation to the subject of inquiry for a substantial period of time; (b) it should be “truthful” and written in “good faith”; (c) it should contain much descriptive material and liberal quotations from those studied; and (d) the procedures for data analysis should be explicit. In addition, Agar (1980: 61) has argued that reports should be written in a style which makes sense to members of the group studied, so that the research may attain later relevance in the context of their own everyday lives.

With more specific reference to the style or rhetoric of scientific accounts, van Maanen (1988), discussing ethnography, identifies two primary types of tales: realist and confessional. The realist tale is the most common and is generally told from the point of view of the subjects of study, with much use of quotations and a focus on everyday life. Confessional tales, instead, stress the field-worker’s point of view and often are intended to explain (and justify) the activities of the researcher. Both realist and confessional tales, however, imply that writing style may be important for the findings that are communicated. This aspect of qualitative research, and more particularly of ethnography, then, “raises the question … whether ethnography (of any sort) is more a science, modeled on standardized techniques and reporting formats, or an art modeled on craftlike standards and style” (van Maanen, 1988: 34). Rock (1979: 21), like van Maanen, has recommended use of literary techniques such as integration of metaphors and analogies in the discourse of research.

In several of these prescriptions, more attention is given to writing style than to the development of theoretical concepts. We submit that both aspects are important for the further development of qualitative research. As a first step, most traditional criteria for research reports also apply to the presentation of qualitative studies, calling for a clearly formulated research problem based on an explicitly stated theoretical perspective; thorough presentation and discussion of relevant literature; adequate elaboration of the chosen methodology; and logical presentation of findings and conclusions. However, because an understanding of the lived experience and everyday reality of research subjects is key not just to the conduct, but also to the appreciation and assessment of qualitative findings, readers should be given an opportunity to relive this experience. Qualitative research findings are constituted through the subjects’ categories of meaning and experience. It is the integration of discursive criteria of scientific reporting with more traditional, substantive criteria which van Maanen characterizes as artistic craftsmanship. One challenge for qualitative social science is to contribute to the current development and clarification of the rhetoric of science (Nash, 1990; Simons, 1989).

There are abundant examples of qualitative sociology which demonstrate artistic craftsmanship. To name a few: Goffman’s (1959; 1963) work on communicative interaction processes; Becker and colleagues’ (1961) examination of medical training and socialization; and the study by Glaser and Strauss (1965) on “leave taking” by patients during the process of terminal illness. All of these studies demonstrate the potential of qualitative methodologies for representing specific social realities. Similar examples from qualitative communication research, while remaining fewer in number, include Epstein’s (1973) News from Nowhere, Tuchman’s (1978) Making News, and Gans’ (1979) Deciding What’s News. It is the task of further studies to integrate such qualities of scientific reporting with an explication of the methodologies employed, the specific approaches to data collection, and, not least, the analytical procedures of the qualitative research process.